We are already seeing how Artificial Intelligence can begin to make creative projects, acting as a form of artificial creativity.
But will it ever be able to truly create a piece of art?
In the interesting video above, host Mike Rugnetta from the PBS Idea Channel discusses his interpretation of whether a computer is able to create something which we would agree is art.
It is a complicated question which brings up some surprisingly nuanced interpretations of what it means for something to be considered art.
And ultimately, while the definition of art will vary between people, communities and over time, it will always require a human to give their approval that it fits with the definition of art.
What is art anyway?
So what is my definition of art? I’m not sure if it is perfect, but I would say it is something along the lines of:
Art is: The skillful and intentional execution of an original idea which conveys a message
If we remember that an idea, especially a creative idea, needs to have both novelty and value, there are a number of components in that definition which all need to work together.
- Skill: Is the artist better at executing their ideas than most people, or produce a more professional / pleasing result? Here, anything which reduces the need for skill, such as Smartphone photo filters, reduces the artistic merit of the result.
- Intentionality: Did the artist have an original idea in mind of what they wanted the end result to be, and worked towards that result? This, combined with skill, usually makes up the “style” of the artist. Here, copying (or just performing) someone else’s idea reduces the artistic merit of the result.
- Message: Does the artist want their work to convey a specific message about the subject beyond the superficial, or illicit a feeling from it? Here, having an output which just represents reality without a deeper meaning reduces the artistic merit of the result. However, this is complicated when other people search for meaning which isn’t there.
The boundaries here will always be blurry, and up for interpretation by various experts and fans. It really is true that beauty is in the eye of the beholder, and the same often goes for art.
After all, there so many examples of artistic pieces which incite fierce debate about whether they should be classified as art at all as it is not clear whether they have a message or skill behind them, including a urinal (Fountain, by Duchamp), a dirty bed (My bed, by Tracey Emin), a blue painting with a white line (Onement VI, by Barnett Newman, which sold for $44 million!) and a dead sheep (Away from the Flock, by Damien Hirst).
But in order for us to ask if computers can produce art, I want to talk specifically about the intentionality of the creation.
And to help illustrate the story, I want to tell you about a young french artist called Pierre Brassau.
In 1964, a new avant-garde artist was introduced to the art scene in the Swedish city of Gōteborg, featuring four paintings in the 1964 exhibition at Gallerie Christinae. The fresh new artist was Pierre Brassau and his work received rave reviews, both from critics and art fans.
He even sold one “masterpiece” to a collector for $90 (about $650 today). The exhibition featured paintings from artists across Europe, but it was the hot new French Artist who stole the show. One critic in particular, Rolf Anderberg, was so overwhelmed by Pierre’s talent that he wrote the following review about his work, which appeared in print the morning following the exhibition:
Brassau paints with powerful strokes, but also with clear determination. His brush strokes twist with furious fastidiousness. Pierre is an artist who performs with the delicacy of a ballet dancer.
The reviews were glowing. All but one. One critic’s review was short and to the point:
All but one.
One critic’s review was short and to the point:
“Only an ape could have done this.”
Little did he know how close to the truth he was. Because this is a picture of Brassau:
The whole thing had been a hoax, orchestrated by tabloid journalist Åke Axelsson to see if critics could actually tell what was good art. Brassau was actually a chimpanzee named Peter from a Swedish Zoo. And the critics had fallen into his trap.
Why do I bring up this story?
Because while the chimpanzee had managed to put paint onto a canvas (after he stopped eating the paint, allegedly Cobalt blue was his favorite), he never set out to intentionally create a specific piece. It was able to pass as abstract work precisely because there was no intention behind it.
And the same is true of computer generated art, just on a much more advanced scale.
So can computers produce art?
We already have numerous software systems capable of generating artistic compositions.
In the video, we learn about Google’s Magenta which aims to create systems which can use neural nets to mirror the way the human brain comes up with solutions. Their aim is stated as a research project to advance the state of the art in machine intelligence for music and art generation.
There are other numerous other examples of creative artificial intelligence, which can produce pictures, soundtracks and even new recipes.
But all of these systems work by scanning huge databases of examples given to them by people, learning what the patterns are, and then using a series of random starting points and the learned pattern to generate new art.
For example, in preparation of the above video about whether a computer could produce art, Mike Rugnetta used similar neural network machine systems to “write” an episode of his show. It analysed hundreds of his previous scripts to find out a flow and style for how he talks, and finally produced this completely digitally scripted episode:
As it may be obvious now, these systems aren’t perfected yet. In the second part of the video, Mike describes how the system worked, and why it produced some strange text. It is clear that while the system was producing text and words, and even a syntax which sounded like sentence flow, the result was barely recognisable as something a human would produce.
You could say it’s artistic skill was very low.
But with increasing amounts of data to experiment with and improved processing power to analyse this data, the ability of such systems to generate creative new content will only improve over time. We truly are still at the infant stages of such creative computers.
However, while the skill level with which these system can produce output will improve (and inevitably surpass humans in some cases), what they lack is the intentionality of an original idea.
They will always be based on a dataset which has been given to them by a group of people. Perhaps their output will be able to evolve over time, much like painting styles changed over time.
Indeed, recently we even saw an example of an AI intentionally playing an original move in the game of Go against a human to win a match, the best example ever of a machine having an idea.
But all of these examples are based on a computer using logic to succeed at a specified challenge it was given by a human, such as “create a melody with this pattern” or “find a Go move which makes victory in the game more likely”.
None of them are examples of a computer producing something because it wanted to.
And this is the difference. Computers will never have the intentionality of creating something, especially not to convey a specific message.
They may be able to generate artistic content, but they will never be able to create art independently of a human giving them instructions.
Finally, the question also stands of whether these new software systems will act as tools for future artists. Whenever a new medium develops, creative people will find ways of using it to express their ideas.
- Think of how the invention of the camera led to the art of photography.
- How radio waves led to widespread listening and demand for popular music.
- How the development of aerosol cans led to graffitti street art.
- How the development of electronic synthesysers and loopers led to dance music.
- And now, the advance of Virtual Reality leading to a whole new medium of video games.
A computer may be able to generate art. A human will then be able to use that computer to create art.
Do you like insights into creativity like this?
Then sign up for your FREE account from Idea to Value to not only get great pieces of insight like this every week, but also free training on improving your creativity and company innovation capabilities from some of the world’s leading innovation experts.
Latest posts by Nick Skillicorn (see all)
- How machines learn (a great video explanation of deep learning) - January 8, 2018
- These Xiaogang farmers could have been executed for their innovation in 1978, instead they changed China forever - December 18, 2017
- Flexing Your (Underused) Creative Muscle: my podcast interview with the Innovation Engine - November 27, 2017
- Why didn’t I think of that? A reusable, square wine barrel wins design award - November 20, 2017